← All comparisons
Claude 4.5 Haiku (Non-reasoning) vs GPT-5.1 Codex mini (high)
Anthropic vs OpenAI — side-by-side benchmark comparison
| Claude 4.5 Haiku (Non-reasoning) | GPT-5.1 Codex mini (high) | |
|---|---|---|
| Intelligence Index | 31.0 | 38.6 |
| Coding Index | 29.6 | 36.4 |
| Math Index | 39.0 | 91.7 |
| Output speed (tok/s) | 105.8 | 218.5 |
| Blended price ($/1M) | $2.19 | $0.69 |
| Time to first token (s) | 0.65s | 3.35s |
| aime | — | — |
| aime 25 | 39.0% | 91.7% |
| artificial analysis coding index | 29.60 | 36.40 |
| artificial analysis intelligence index | 31.00 | 38.60 |
| artificial analysis math index | 39.00 | 91.70 |
| gpqa | 64.6% | 81.3% |
| hle | 4.3% | 16.9% |
| ifbench | 42.0% | 67.9% |
| lcr | 43.7% | 62.7% |
| livecodebench | 51.1% | 83.6% |
| math 500 | — | — |
| mmlu pro | 80.0% | 82.0% |
| scicode | 34.4% | 42.6% |
| tau2 | 32.5% | 62.9% |
| terminalbench hard | 27.3% | 33.3% |
Benchmark data from Artificial Analysis.